N4-fields: Neural network nearest neighbor fields for image transforms

Yaroslav Ganin, Victor Lempitsky

    Результат исследований: Глава в книге, отчете, сборнике статейМатериалы для конференциирецензирование

    52 Цитирования (Scopus)

    Аннотация

    We propose a new architecture for difficult image processing operations, such as natural edge detection or thin object segmentation. The architecture is based on a simple combination of convolutional neural networks with the nearest neighbor search. We focus our attention on the situations when the desired image transformation is too hard for a neural network to learn explicitly. We show that in such situations the use of the nearest neighbor search on top of the network output allows to improve the results considerably and to account for the underfitting effect during the neural network training. The approach is validated on three challenging benchmarks, where the performance of the proposed architecture matches or exceeds the stateof- the-art.

    Язык оригиналаАнглийский
    Название основной публикацииComputer Vision - ACCV 2014 - 12th Asian Conference on Computer Vision, Revised Selected Papers
    РедакторыMing-Hsuan Yang, Hideo Saito, Daniel Cremers, Ian Reid
    ИздательSpringer Verlag
    Страницы536-551
    Число страниц16
    ISBN (печатное издание)9783319168074
    DOI
    СостояниеОпубликовано - 2015
    Событие12th Asian Conference on Computer Vision, ACCV 2014 - Singapore, Сингапур
    Продолжительность: 1 нояб. 20145 нояб. 2014

    Серия публикаций

    НазваниеLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    Том9004
    ISSN (печатное издание)0302-9743
    ISSN (электронное издание)1611-3349

    Конференция

    Конференция12th Asian Conference on Computer Vision, ACCV 2014
    Страна/TерриторияСингапур
    ГородSingapore
    Период1/11/145/11/14

    Fingerprint

    Подробные сведения о темах исследования «N4-fields: Neural network nearest neighbor fields for image transforms». Вместе они формируют уникальный семантический отпечаток (fingerprint).

    Цитировать